Google unveils Gemini, its most powerful AI model to date

Google’s Gemini will appear in different forms and should outperform GPT-4. In this way, Google wants to prove that it, and not anyone else, is the AI ​​company par excellence.

Sundar Pichai (CEO, Google) and Demis Hassabis (CEO, Deepmind) hiring Gemini proudly. Gemini is the new, cutting-edge LLM certification that has come out of Deepmind Labs. Just a few days ago, it looked like Gemini wouldn’t see the light of day until 2024, but Google can’t wait to make that announcement. With Gemini, Google wants to once again show the world that it is an “AI-first” company, a reputation that has somewhat crumbled since the launch of ChatGPT. Gemini appears in three forms: Super, forefront And Nano.

Google isn’t shy about comparing it to GPT-4. One blog lists 32 benchmarks, of which the Gemini Ultra scores better than the competition, which has thirty benchmarks. The Gemini Ultra is also said to have achieved a score of ninety percent in multitasking tests that combine different subjects. This makes Gemini the first AI model to score better than human experts on these tests. The AI ​​model is also capable of providing more complex reasoning.

To train Gemini, Google has also developed a new generation of TPU for its data centers: Cloud TPU v5p. Despite the expanded capabilities, applications built on Gemini models can run faster than their less “heavy” counterparts.

Great upgrade for cool

The theory sounds great, but what about practice? As a regular human, you have the best chance of connecting with Gemini Pro. Google wants to roll out the mid-version of Gemini widely in its own ecosystem, starting with cold, which is getting its biggest update yet. Google launched Bard in February to compete with ChatGPT, but the chatbot has never been able to match that success. That has to change with Gemini.

Read also

One year of ChatGPT: unprecedented hype, unprecedented profits, and unprecedented behind-the-scenes TV drama

Thanks to Gemini, your chatbot will have better reasoning abilities, leading to better and more relevant answers. But Bard will also increase programming knowledge and expand it with multimedia capabilities. The latter means that Bard is not limited to text-to-text conversion, but can also process and produce images, audio, and video.

Google is showing a demo of how a well-known influencer made a paper airplane with the help of Bard via its sister YouTube channel. We note that these functions are also included in the latest version of ChatGPT. Therefore, it has become easier to follow what Google is doing. The new Bard capabilities are currently only available in English in 170 countries as of today.

Google plans to launch a “Bard Advanced” version early next year that will run on the Gemini Ultra. Gemini Pro will also be gradually introduced within Search, Chrome, and other Google services. The model will also be offered to developers in Google Cloud via Vertex AI.

Artificial intelligence on your smartphone

To make sure you get the best that today’s AI has to offer, always and everywhere, Google is launching Gemini Nano. This is a “smaller” version of LLM that was specifically designed for use in smartphone processors. Google has actually prepared its latest Pixel smartphones for this purpose by putting TPU material in the Tensor chip. As a result, many new AI functions, including image processing, will be rolled out to the latest devices with software updates.

The more powerful version, Gemini Ultra, will remain in the waiting room for a while. Only a limited number of customers are allowed to try out the model in a controlled environment. Gemini Ultra takes a big step toward artificial general intelligence, where AI is quite competitive with humans, so Google wants to conduct more extensive security testing before putting the technology in the hands of the masses.

However, with the launch of Gemini, Google shows that it is once again fully involved in the AI ​​race. It’s clear that the hype around generative AI won’t die down right away in 2024. While the focus this year was mainly on play and experience, the added value of the technology should really become visible next year.